<no title>

	Year	Salary	JobSat	YearsCode	YearsCodePro	Age	Education	OrgSize	LastNewJob	Employment	RespondentType	JobSeek	Gender	Student	Country	CodingActivities	DevType	LearnCodeFrom	LangPresent
0	2022	69318.0	<NA>	10	5	25-34	master	500 to 999 employees	<NA>	fulltime	dev	<NA>	male	no	Germany	School or academic work	Data scientist or machine learning specialist;...	Books / Physical media;School (i.e., Universit...	C;C++;Java;JavaScript;MATLAB;Python;Scala;SQL;...
6	2022	27652.0	<NA>	18	10	25-34	bachelor	1,000 to 4,999 employees	<NA>	fulltime	dev	<NA>	male	no	Colombia	Hobby	Developer, full-stack;Developer, back-end	Books / Physical media;Other online resources ...	Bash/Shell/PowerShell;Elixir;HTML/CSS;JavaScri...
9	2022	15431.0	<NA>	5	5	25-34	bachelor	20 to 99 employees	<NA>	fulltime	dev	<NA>	male	no	Ghana	Freelance/contract work	Developer, back-end	On the job training;Coding Bootcamp	JavaScript;Ruby
13	2022	47352.0	<NA>	7	7	45-54	master	10 to 19 employees	<NA>	fulltime	non-dev	<NA>	male	no	Belgium	Hobby	Developer, back-end;Educator or academic;Datab...	Books / Physical media;On the job training;Col...	Delphi;SQL
22	2022	78084.0	<NA>	25	25	45-54	bachelor	500 to 999 employees	<NA>	fulltime	non-dev	<NA>	male	no	Canada	Hobby;Contribute to open-source projects	Engineer, site reliability;Security professional	Books / Physical media;Other online resources ...	Bash/Shell/PowerShell;C;JavaScript;Perl;PHP;Py...

# Show the top-n countries in the plot.
TOP_N_COUNTRIES = 10

# Take the top-n countries that have the most developers.
countries = df[['Country', 'RespondentType']] \
    .query('RespondentType == "dev"') \
    .query('Country != "Brazil"') \
    .query('Country != "India"') \
    .query('Country != "Russia"') \
    .query('Country != "Sweden"') \
    .groupby('Country', as_index=False) \
    .count() \
    .sort_values('RespondentType', ascending=False) \
    .head(TOP_N_COUNTRIES) \
    ['Country'] \
    .tolist()

countries_df = df[df['Country'].isin(countries)]
    
# Calculate the salary gap in percentage.
salary_men_df = countries_df.query('Gender == "male"') \
    .groupby(['Year', 'Country']) \
    .agg({'Salary': 'mean' }) \
    .reset_index() \
    .rename(columns={'Salary': 'SalaryMen'})

salary_women_df = countries_df.query('Gender == "female"') \
    .groupby(['Year', 'Country']) \
    .agg({'Salary': 'mean' }) \
    .reset_index() \
    .rename(columns={'Salary': 'SalaryWomen'})

salary_gap_df = pd.merge(salary_men_df, salary_women_df, on=['Year', 'Country']).dropna()
salary_gap_df['SalaryGap'] = abs(salary_gap_df['SalaryMen'] - salary_gap_df['SalaryWomen'])
salary_gap_df['SalaryGapPercent'] = (salary_gap_df['SalaryMen'] - salary_gap_df['SalaryWomen']) / salary_gap_df['SalaryMen'] * 100

salary_gap_df.head()

	Year	Country	SalaryMen	SalaryWomen	SalaryGap	SalaryGapPercent
0	2014	Australia	81355.140187	70000.0	11355.140187	13.957496
1	2014	Canada	71124.260355	57692.307692	13431.952663	18.885191
2	2014	France	54724.409449	47000.0	7724.409449	14.115108
3	2014	Germany	61491.596639	50714.285714	10777.310924	17.526478
4	2014	Italy	33923.076923	25000.0	8923.076923	26.303855

# Calculate the mean salary gap for all countries for all genders.
mean_salary_men_df = countries_df.query('Gender == "male"') \
    .groupby(['Year', 'Country']) \
    .agg({'Salary': 'mean' }) \
    .reset_index() \
    .rename(columns={'Salary': 'SalaryMen'})

mean_salary_women_df = countries_df.query('Gender == "female"') \
    .groupby(['Year', 'Country']) \
    .agg({'Salary': 'mean' }) \
    .reset_index() \
    .rename(columns={'Salary': 'SalaryWomen'})

mean_salary_gap_df = pd.merge(mean_salary_men_df, mean_salary_women_df, on=['Year', 'Country']).dropna()
mean_salary_gap_df['SalaryGap'] = abs(mean_salary_gap_df['SalaryMen'] - mean_salary_gap_df['SalaryWomen'])
mean_salary_gap_df['SalaryGapPercent'] = (mean_salary_gap_df['SalaryMen'] - mean_salary_gap_df['SalaryWomen']) / mean_salary_gap_df['SalaryMen'] * 100

mean_salary_gap = mean_salary_gap_df.groupby('Year', as_index=False)['SalaryGapPercent'].mean()

mean_salary_gap

	Year	SalaryGapPercent
0	2014	25.065
1	2015	23.649814
2	2016	18.958623
3	2017	14.428017
4	2018	14.927359
5	2019	16.585776
6	2020	18.866154
7	2021	21.314563
8	2022	16.113038

color_alpha = 0.7

fig = px.line(
    salary_gap_df,
    x='Year',
    y='SalaryGapPercent',
    color='Country',
    hover_name='Country',
    title='Top-10 Country Salary Gap Between Men and Women<br><sup>Generally decreasing, but still male-favoured</sup>',
    custom_data=['Country'],
    color_discrete_sequence=[
        f'rgba(233, 146, 119, {color_alpha})',
        f'rgba(148, 233, 119, {color_alpha})',
        f'rgba(119, 127, 233, {color_alpha})',
        f'rgba(218, 119, 233, {color_alpha})',
        f'rgba(218, 191, 162, {color_alpha})',
        f'rgba(207, 150, 150, {color_alpha})',
        f'rgba(167, 199, 213, {color_alpha})',
        f'rgba(233, 119, 186, {color_alpha})',
        f'rgba(154, 162, 221, {color_alpha})',
        f'rgba(154, 221, 209, {color_alpha})',
    ],
    range_y=[0, 48],
    width=790,
)

fig.update_layout(yaxis_title='Salary gap in percentage')

# Add male/female ratio in the legend names
total_male_participants_per_country_df = countries_df.query('Gender == "male"') \
    .groupby('Country', as_index=False) \
    .size() \
    .rename(columns={'size': 'TotalMale'})

total_female_participants_per_country_df = countries_df.query('Gender == "female"') \
    .groupby('Country', as_index=False) \
    .size() \
    .rename(columns={'size': 'TotalFemale'})

total_participants_per_country_df = pd.merge(total_male_participants_per_country_df, total_female_participants_per_country_df)
total_participants_per_country_df['MaleRatioPerFemale'] = total_participants_per_country_df['TotalMale'] / total_participants_per_country_df['TotalFemale']

# Add female/male ratio for each legend name.

# Example: {'oldname': 'newname'}
legend_names = {}
for _, row in total_participants_per_country_df.iterrows():
    legend_names[row['Country']] = f"{row['Country']} (1/{int(row['MaleRatioPerFemale'])})"

fig.for_each_trace(lambda t: t.update(name=legend_names[t.name], 
                                      legendgroup=legend_names[t.name],
                                      hovertemplate='<b>' + t.name + '</b><br>Year: %{x:d} </br>Salary gap: %{y:d}%<extra></extra>'))

# Plot the mean salary scatters
mean_salary_line_fig = px.line(mean_salary_gap,
                               x='Year',
                               y='SalaryGapPercent',
                               markers=True)
for trace in mean_salary_line_fig.data:
    trace.hovertemplate = '<b>' + trace.name + '</b><br>Year: %{x:d} </br>Mean salary gap: %{y:d}%<extra></extra>'

fig.add_traces(list(mean_salary_line_fig.select_traces()))
fig.add_annotation(x=2019.7, y=17.75,
                   showarrow=False,
                   xanchor='center',
                   yanchor='bottom',
                   textangle=-16,
                   text='mean salary gap')

# Add caption
fig.add_annotation(x=-0.11, y=-0.35,
                   xref='paper', yref='paper',
                   showarrow=False,
                   xanchor='left', yanchor='bottom',
                   align='left',
                   text='Salary gaps across ten of the countries with the most survey respondents.<br>Positive percentages indicates a male-favoured salary gap.')

fig.update_layout(
    legend={
        'title': 'Country with<br>female/male ratio<br>'
    },
    margin={'b': 130, 't': 100, 'l': 90}
)

fig.show()

Contents